Translating the ECB's phased deployment schedule into the specific architectural and operational obligations that settlement system builders must satisfy before each phase gate closes.

*European agentic commerce is building its reasoning layer on US-hosted infrastructure, and the terms of that dependency have not yet been set by European law, European operators, or European capability.*

*A layer-by-layer control assessment of where enforceable EU obligations and user-governance mechanisms hold, falter, or are structurally absent across the agentic commerce architecture.*

*An examination of the structural mechanisms (consent portability, sector-specific connectors, and regulatory alignment) that position European federated architectures ahead of centralised alternatives on measurable governance and market-access grounds.*

*A taxonomy of buildable agent systems rated against data residency, open-source governance, and integration with EU payment and identity infrastructure, where framework choice carries direct regulatory consequence.*

*Five European jurisdictions are pulling ahead in agentic commerce, and the structural conditions separating leaders from laggards are more specific than regulatory friendliness alone.*

*The European Digital Identity Wallet enters its mandated rollout window as agentic commerce demands credential and delegation semantics the current architecture has not yet been designed to carry.*

*Brussels, Washington, and London are writing incompatible legal definitions of AI-driven autonomous commerce, and the jurisdiction that resolves the liability question first will set the architecture standard for all others.*

*The vocabulary of classical EU agency law and the vocabulary of automated commerce are converging on the same five terms, carrying different meanings and distinct liability consequences.*

*As EU legislation, autonomous agent architectures, and cross-border case law converge on the same transactions, the vocabulary used to describe them is fracturing, and the cost of that fracture is no longer theoretical.*

*PSD3 and its accompanying Payment Services Regulation arrive with authorization frameworks designed for human decision-makers, at the precise moment that agentic commerce requires something structurally different.*

*Four structural criteria now determine whether a commerce system carries scripted-automation accountability or the elevated liability profile of an autonomous agent under European regulatory frameworks.*

*Where European commerce platforms place the boundary between human discretion and machine execution is a regulatory and operational design choice that existing EU law was not written to adjudicate end-to-end.*

*Where autonomous commercial agents sit within the EU AI Act's risk taxonomy remains unsettled, yet the compliance obligations that attach to their core behaviours are already active.*

*Two regulatory architectures have settled on opposite answers to the same question: when, and at what frequency, must an issuer show its reserve to the world.*